Open Myoelectric Signal Processor

OPP is seeking to encourage experimentation with myoelectric control in order to inspire more rapid development of mechatronic prostheses for amputees. Because of the extremely small market size for upper extremity prosthetics, we think that one way to encourage this activity is to develop toys or user customizable devices that are capable of myoelectric signal processing. With wider access and experimentation with the technology, perhaps we could see interesting developments beyond the traditional venues of corporate R&D and educational institution research.

Along with a couple of collaborators (and we'd love to have more), we are exploring the development of an extremely flexible open hardware module that could be used with the Buglabs device, the LEGO NXT platform, and as a USB device that could be used with the OLPC (laptop.org), and traditional linux and windows computers. The components of the device would be modular, so we could populate them as needed, and use the same design for a variety of applications. The USB interface could also be used for connecting to game consoles, such as the Xbox360 and Playstation 3; the I2C bus may be useful for connecting to the Nintendo Wii via the Nunchuck input.

This project is in the very early stages of development, and we intend to use this space for the initial exploration of the project, including the development of a problem statement and design requirements for the device. We are planning on running all electronic source documents, including the board schematics, through a sourceforge-hosted site. Please contact us if you are interested in contributing.

Source for hardware and software will be maintained with version control, etc., on our Google Code page. The project name, created in a punny and hopefully not disastrously nearsighted moment, is "MyOpen." Strenuous objections can be raised in any available forum.

Design a device capable of collecting 16 or fewer channels of skin surface myoelectric signals (EMG), processing the collected data, and delivering to a host device the results of signal processing algorithms such as pattern recognition in real time.

The goal is to make an affordable, safe, open-source EMG processor for the following markets:

• Prosthetic users
  

• Gamers
  

• Education, e.g. http://www.vernier.com/nxt/index.html
  

• Hackers, students, creators, and other innovators.

We would like the device to be

• Available for approximately $250
  

• Available for cheaper in kit-form
  

• capable of a variety of modes of communication with a host device, including
  

• I2C / TWI
  

• Ethernet
  

• USB
  

• Analog (e.g. onboard DAC)
  

• Otto Bock prosthetics require 0-5V signaling, offset 1.25/1.3V.
  

• May also want to add analog feedback to user, e.g. through vibrator motors.
  

• JTAG
  

• RS-232 (?)
  

• CAN (?)
  

• Bluetooth? (e.g. via BlueCore-4 in the NXT brick)
  

• Battery or USB powered ( 3 - 5.5V in)
  

• hence, low power, < 150mw. Each interface must be able to be switched off.

http://openeeg.sourceforge.net/doc/
https://sourceforge.net/projects/oip

From the MFAD program at SVA that is currently working on prosthetics from a Graphic Design and 3D form approach:
It struck me when Jon Kuniholm spoke to us that there is a need to develop MyoElectric Technology further and bring down the price. I think there would be a lot of interest in this technology in the private market for other uses, that could serve to further software, hardware and cost if it was attached to the right products. Video games have obvious applications: who can forget the power glove and imagine if it was actually a glove that reacted to your muscle movement! Attached is a clip of someone who got his hands on some of the MyoElectric sensors and worked out some software to compose music. I think this is amazing: http://www.youtube.com/watch?v=dhRJqZ33_3I. I think if this technology was opened up to a mass market we would see changes happen quickly, it may just be a matter of getting the right people interested. If anyone has questions or comments please send me an email at mattluckhurst@yahoo.com. Please feel free to examine this idea further and let me know if it is feasible or not. Thanks.

We had originally decided to develop both hardware and software under the Lesser General Public License (L-GPL). We made this decision because we wanted to allow the possibility for commercial prosthetic devices to link proprietary software to the software we are releasing, as well as create hardware that does the same thing (need more detail on L-GPL implications for circuit schematics).

We have revisited this decision, and the project is now GPL.

We have been accused of being naive for thinking that we can get away without addressing the issue of intellectual property, and that we need to at least come up with some sort of license for what IP folks believe must be out there already covering this area. This patent covers potentially related idea, referencing only the wireless capability. Volunteers to help do a real patent search and begin negotiating away whatever roadblocks there are would be greatly appreciated.

Prosthetic components are regulated by the federal government as Class I (exempt) medical devices, as discussed here. As the discussion points out, a device marketed as a prosthetic component is subject only to the subset of general requirements that involve record-keeping, complaint tracking, and establishment registration. Because a myo signal processor capable of pattern recognition operates on the same fundamental scientific principle, i.e. skin surface EMG signals, it would not be subject to any exception to exemption described in section 21CFR90.9, just as the example given in the regulation would not deny an exemption to a cutting device that cut by the same method, e.g. blade as opposed to laser. Given the apparent ease with which these requirements could be met, it might be easier to market this device as a prosthetic component than nearly anything else, including a toy.l

Other regulations are a thorny issue - things like UL approval will most likely be costly ($30,000), but will be necessary if we were to sell hardware products directly to consumers. Certifying the equipment will necessarily add to the cost of the device, as will insuring the manufacturers against liability (large as the device will be electrically connected to the user). We are currently looking for ideas to mitigate potential liability pitfalls, one idea is to offer the devices only in kit or incomplete form. Additionally, we should try to meet IEC 60601-1 criteria established for medical devices. The hardware should be designed such that it is possible to certify it if need be - and, besides, these regulations are meant to enforce product safety, and we do want to make the system as safe as possible! At present, the safety requirements include:

• 4kv user isolation from mains and / or computer interface (USB, ethernet, etc)
  

• Robust EMI protection on the input, via transistors, TVS, or diodes.
  

• Open-circuit failure mode for any amplifier inputs. I've been in an OR where the electrophysiology amplifiers (Plexon) failed to one of the +-2.5 power rails, and I do not want to see that again.
  

• Predictable failure modes for any of the other inputs to prevent excessive current input (e.g. USB somehow connected to mains) from being reflected in the isolated power supply.
  

• Robust to strong RF interference. This may be just a matter of proper shielding & software rejection of faulty signals.

In order of priority:

• Lego NXT (I2C interface .. but beware, not open collector)
  

• Buglabs
  

• SPI / I2C, details forthcoming.
  

• To connect as a buglabs device, we'll need to fit their size requirements. At present this is estimated to be 2" x 2" x 1/2".
  

• Wii Nunchuk remote (I2C interface, slightly encrypted but still hackable)
  

• If we want to fully replace the Nunchuck, it will be necessary to add a 3-axis accelerometer.
  

• Windows / Linux computer. Through either USB or Ethernet. (Ethernet is preferred, as driver development is much easier).
  

• Xbox 360 / Playstation 3 (through USB, though driver/software development there may be challenging)